A Program for Minimization of the Effects of Interference on GBT Observations R. Fisher, June 1996 Summary This memo suggests a long range plan for minimizing the effects of interference on GBT observations. The immediate needs of the GBT are efficient identification of interference sources, creation and maintenance of interference spectrum data bases, and effective user displays. Our ultimate goal is to remove interference from astronomical data using sidelobe cancelling, polarization discrimination, better receiver filtering and dynamic range, and new data analysis procedures. Post-observation data processing is, at best, only partially effective at removing interference from astronomical data, so strong emphasis needs to be put on real-time interference cancelling techniques. This requires research and development in a number of areas. First, enhance our ability to identify and measure the properties of interference arriving at the telescope. Second, develop an accurate model of the GBT antenna sidelobes through a combination of theoretical models and pattern measurements. Third, develop techniques for efficiently identifying interference and generating nulls on this interference somewhere in the time, frequency, and spatial domain. Some of this is beyond the current state of the antenna and electronics arts, but the techniques familiar to radio astronomy are uniquely suited to the tasks. The initial instrumentation needed to begin the interference identification and measurement phase include: 1. Several wideband direction-finding and polarization measuring antennas. The 0.1-1 GHz dual polarized log periodic from the antenna range and a 4-antenna log periodic array constructed in the shop would be a good start. 2. Programmable antenna rotator. 3. Two-channel 0.1-2 GHz receiver with a common programmable LO. 4. Spectral processor for signal correlation and spectrum analysis. When the GBT spectrometer is completed, the spectral processor can be the primary interference analysis system at the GBT. Until the GBT control room is completed, interference measurements can be made from the 140-ft with antennas mounted on the roof of the service tower. Eventually, the monitoring and direction-finding antennas will be mounted on a gimbal on top of the GBT feed arm. Some measurement of interference signal properties will require more elaborate antennas than can be mounted on the GBT, so a separate area that is near the GBT control room and high enough to see over the trees will be needed. The frequency range of the antennas and receivers can be extended as experience and requirements dictate. Introduction In principle, if interference is not coincident in time, frequency, polarization, and direction with a celestial object, we should be able to observe in the presence of the interference. We have generally relied on separation of celestial signals from interference in the frequency domain, but this is already inadequate below 1.4 GHz and is becoming less tenable at higher frequencies. Greater separation in the time, spatial, and polarization domains are required. The offset reflector design of the GBT reduces its far sidelobe level by reducing the scatter from the antenna structure, but it will still require the same sidelobe cancelling measures that could be applied to any antenna. The electronics literature is rich in techniques for adaptively cancelling sidelobes, but radio astronomy antennas will require substantial enhancement of these techniques because of our weak signal and high sensitivity requirements. There isn't enough information in the main signal channel combined with auxiliary interference receptors to completely cancel the interference and without loss of sensitivity. Hence, we need to independently measure the spatial properties of each interfering signal and to understand the GBT sidelobe structure well enough to construct nulls in known directions in real time without sole benefit of the signal in that direction. A much better theoretical and empirical model of the GBT far sidelobe pattern than is currently available will need to be developed. A big bonus to understanding the far sidelobe structure will be the suppression of spectral baseline ripples due to the sun and other strong continuum sources. The time resolution for discrimination of interference in spectral line observations is typically tens of seconds. This must be pushed to fractions of a second or even to microseconds. Each interfering source has its own duty cycle and intensity variability time scale which must be resolved to effectively excise it from the data. In many cases, a separate monitor or prediction of the interfering signal's intensity as a function of time will be required. The intensity distribution of weak interfering signals is frequently nearly gaussian, which makes it very difficult to separate from the data after the observations are recorded. More post-observation analysis techniques for flagging and excising interference do need to be developed, but this cannot be the dominant thrust of an interference reduction program. We understand very little about the received characteristics of over-the-horizon interfering signals. Some recent observations of the polarization properties of interference at the 140-ft indicate that multi-path delays exist in many signals. This implies that there may be more than one direction of arrival for many signals and that their time domain properties are considerably modified by propagation to the GBT. Before we can design high-suppression nulls in the time and spatial domain, we must understand more about the properties of received signals. Interference measurement antennas and receivers need to be designed for this purpose. We must not forget that we are our own greatest source of interference. The first step in any comprehensive interference program is to identify and, where possible, eliminate all local signals that can enter the receiver system. Each receiver component and system should be tested for spurious spectral features. Each potential source of interference such as computers, test equipment, motors, and power switching equipment must be measured for radiation. A continuously updated spectral record of the telescope control room and selected sites around the GBT must be recorded and made readily accessible to observers and interference search people. Finally, interference information and suppression techniques will be effective only if they are presented to the users in an understandable and readily implemented way. Data base management and the creation of data displays will be a significant part of the interference suppression program. Interference Signal Properties We are not generally interested in the properties of interfering signals for their own sake, but we do need to know them to design effective countermeasures. Here is a list of properties that should be either measured or obtained from licensing agencies or owners. 1. Polarization matrix vs time. This is signal intensity as a function of time including polarization information. Temporal information might be obtained from the nature of the transmission, e.g. radar rep rate, satellite switching cycle, etc., but the polarization properties must be measured. We would like to know whether we can suppress the signal by blanking when it is on or by extracting its polarized component from the data. Are its intensity statistics sufficiently non-gaussian to eliminate it with post-processing algorithms? 2. Direction vs time. Is the direction of arrival stable or predictable enough to design an effective cancelling scheme? Does the source appear to be extended by multi-path or other scattering? Is this pattern stable, and, if not, what is its variability time scale? We are interested in angular scales as small as a few percent of the GBT Beamwidth. 3. Frequency and bandwidth. What is its spectrum, and how does it vary with time? Are there times when the signal is unmodulated? Is there intensity correlation between different parts of the spectrum that would allow, for example, sidebands to be subtracted on the basis of carrier strength? 4. Emission type (TV video, FM, pulsed, spread spectrum, etc.) 5. Transmitted polarization 6. Transmitter location (Longitude and Latitude in the case of terrestrial signals or orbit coordinates in the case of spacecraft.) 7. Transmitted power and predicted path loss 8. Transmitter owner and other administrative information Signal Measurement Equipment For most interference measurements a conventional spectrum analyzer is not sensitive enough, and it cannot make polarization measurements. During the time that the spectral processor is not in use for astronomical measurements, it will make an ideal interference analyzer. To it we need to add a moderately sensitive dual-channel front-end with an IF bandwidth of about 50 MHz and a common LO under computer control. Eventually, we should have a receiver and antennas that cover all of the GBT receiver frequencies for which we expect interference, but we can start with the 0.1-1.7 GHz frequency range where the interference is worst. For monitoring and source identification three antenna systems are required: 1. a well matched and directive portable antenna for control room, selected site, and equipment radiation measurements; 2. a 360-degree direction-finding antenna for signals near the horizon; and 3. a broad-beam, vertically directed antenna for satellite monitoring. The latter two should be located reasonably close to one another so that they receive roughly the same signal strength and so that they can be selected in quick succession under computer control. Many signals vary in strength over short periods of time, so the direction finding antenna must measure directions instantaneously. Mechanical antenna rotation is too slow as a primary method of direction finding. Antennas and receivers for measuring interference signal properties will be developed as experience suggests. We will start with a wideband dual polarized antenna for polarization measurements. Both the monitoring and research antennas should have good internal and external intensity calibration standards for accurate field strength measurements. For rapid, very broad band monitoring and spectrum recording a general purpose digital spectrum analyzer is required. For time resolutions less than 100 microseconds we need a total power detector and oscilloscope, preferably a digital recording instrument. The spectrum analyzer and oscilloscope can serve as backups for monitoring when the spectral processor is in use for astronomy. A significant software effort will be required for programming the spectral processor for interference monitoring, for control of the interference receiver LO, programming the spectrum analyzer, and for developing data analysis, display, and archiving procedures. GBT Antenna Properties Strong cancellation of antenna sidelobes in the direction of interference will require an accurate and detailed understanding of the GBT response pattern in all directions and frequencies under all configurations such as different receiver focus and subreflector positions. A straightforward calculation or measurement of the pattern would require totally impractical amounts of time and effort, and the techniques for calculating the pattern to required accuracy are beyond the current state of the art. Combined theoretical and empirical techniques aimed specifically at the sidelobe cancelling problem need to be developed. The likely approach will be to create efficient display and analysis methods for comparing the best current computed and measured antenna patterns, understand the differences and improve the theory to account for more detail, improve the efficiency of measuring the antenna pattern using low-orbit satellite beacons, for example, and develop methods for interpolating and extrapolating antenna patterns from patterns sampled sparsely in direction, frequency, and antenna configuration. The use of improved far sidelobe information need not wait for the accuracy required for sidelobe cancelling. Astronomical data can benefit from the avoidance of observations when the strongest sidelobes are pointed at known interference sources. Also, advantage might be taken of existing nulls in the sidelobe pattern. A fairly sophisticated display and scheduling system will be of great benefit in this regard. User Displays The effectiveness of any interference countermeasures will depend heavily on presenting the information to the observer and telescope operator. Most countermeasures will require human decisions. The displays must present the information quickly, provide a history of the interference environment, and present as much information as can be contained on one display without confusion. Some display examples are: 1. Direction finding maps on various horizontal scales from the local site to about a 500 km radius. Important landmarks, geographic information (cities, roads, etc.), and potential sources of interference need to be on this display. 2. Grey scale of intensity as a function of time and frequency. The time axis could be quite long and require a scrolling facility. 3. Grey scale of intensity as a function of two frequencies. For example, we might want a plot of the IF passband spectrum as a function of an LO frequency. 4. Intensity vs frequency for a given time or time range. 5. Intensity vs time for a given frequency or passband for time scales from microseconds to many days. 6. Intensity histogram for a given frequency or passband and given time interval. 7. The antenna pattern with respect to known sources of interference, both man-made and celestial, and with respect to the ground. Data Bases Current instances of interference need to be interpreted in the context of the interference history and of the known sources of interference in the same frequency range. An interference monitoring history of a specified frequency range is helpful to observers planning future observations and to frequency allocation people for making a case for interference protection. The following data bases are needed: 1. Receiver passband spectra in the absence of antenna input. At some level every receiver generates its own spurious signals. The intrinsic receiver spectrum should be saved in a data base to reduce the possibility that it will be confused with external interference. In the course of GBT development every receiver component should be examined for spurious signals and corrected to the extent possible. All components following the first receiver amplifier should be tested with an input noise level roughly 20 dB below its nominal input to expose the spurious signals in a 1-minute spectrum integration. The data base should include spectra taken under as many combinations of LO frequencies and other operational parameters as practical and useful. 2. Local radiation environment. Each potential source of interference in the vicinity of the GBT should be examined for radiation and its spectrum entered in the data base. Broadband spectra from selected locations in the control room and in the vicinity of the GBT, including the top of the GBT feed arm, should be archived along with the locations and details of the measurements. 3. Locations of known and potential interference sources with the information listed in the section "Interference Signal Properties." 4. Terrain and geographic data for the observatory site vicinity, the National Radio Quiet Zone, and roughly a 500 km radius around the observatory. The terrain information outside the NRQZ can be of lower resolution. 5. The GBT antenna sidelobe pattern (measured and computed) for as many frequencies as practical. Data Analysis Tools Data analysis tools are required both for interference signal analysis and for excising interference from astronomical data. Some examples of the latter that have proven to be of value are: 1. Detection and removal of non-gaussian distributed instances of interference intensity. 2. Removal of signal sidebands on the basis of their correlation with the carrier. 3. Spectrum cleaning using the known spectrometer response function. 4. Removal of polarized interference flux. 5. Data flagging and quality assessment on the basis of polarization, intensity statistics, and adjacent beam comparison.